GNU bug report logs - #44131
28.0.50; emacsclient Invalid socket owner error

Previous Next

Full log

View this message in rfc822 format

From: Eli Zaretskii <eliz <at> gnu.org>
To: Andrew Eggenberger <andrew.eggenberger <at> gmail.com>, Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 44131 <at> debbugs.gnu.org, akrl <at> sdf.org
Subject: bug#44131: 28.0.50; emacsclient Invalid socket owner error
Date: Sat, 07 Nov 2020 10:47:35 +0200

> From: Andrew Eggenberger <andrew.eggenberger <at> gmail.com>
> Date: Sun, 1 Nov 2020 19:24:42 -0600
> Cc: 44131 <at> debbugs.gnu.org
> 
> The problem seems to stem from the way the hurd treats fstat calls on local socket
> file descriptors. They all have the uid of 0 (root), while emacsclient can be run by 
> other users. After trying to teach the hurd's local socket server to use the uid of 
> the user who requested the socket and failing, I created the naive patch below that
> fixes the problem.
> 
> Please let me know if there's a better solution. I git blamed the changes that led to this
> issue and I don't understand the race condition the uid comparison is supposed prevent.
> 
> Andrew Eggenberger
> 
> diff --git a/lib-src/emacsclient.c b/lib-src/emacsclient.c
> index 871fa7a8d3..6059993ff6 100644
> --- a/lib-src/emacsclient.c
> +++ b/lib-src/emacsclient.c
> @@ -1480,8 +1480,13 @@ set_local_socket (char const *server_name)
>         sock_status = errno;
>        else if (connect_stat.st_uid == uid)
>         return s;
> +#ifdef __GNU__
> +      else
> +       return s;
> +#else
>        else
>         sock_status = -1;
> +#endif
> 
>        CLOSE_SOCKET (s);
>      }

Paul, any better ideas?  Or should I push this fix?

Thanks.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.